Selective ethylene trimerization: A study into the mechanism and the reduction of PE formation

Article abstract of DOI:10.1016/j.molcata.2005.12.030

In the present study, the titanium-catalyzed ethylene trimerization in general, and more specifically, the concomitant PE formation have been studied. The polymer formation is undesirable as it not only will lead to lower 1-hexene yields, but it can cause reactor fouling under the applied conditions (30-80 °C, toluene solvent). It is, therefore, important to know which factors are involved in the formation of polymeric products and how their formation can be reduced or even prevented. The PE formation turns out to be catalyzed by at least two different species. A significant amount of PE is formed in the early stages of the reaction, caused by the presence of partly alkylated titanium species. The PE formation during later stages of the reaction is due to degraded catalyst species, which means that polymer formation is an inherent property of this catalyst system. The polymer output can be reduced largely by premixing Cp′TiCl₃ and methyl aluminoxane (MAO) prior to injection into the reactor. It was also demonstrated that the type of MAO activator/impurity scavenger is of great importance. For a low yield of PE it is essential to use an MAO that does not contain and/or is not able to generate aluminum hydride species. In the end the best results with respect to both productivity and selectivity were obtained by starting from trimethyl titanium compounds.

Full text of DOI:10.1016/j.molcata.2005.12.030

Journal of Molecular Catalysis A: Chemical 248 (2006) 237–247
Selective ethylene trimerization: A study into the mechanism
and the reduction of PE formation
Henk Hagena,∗, Winfried P. Kretschmerb, Frederik R. van Burena,
Bart Hessenb, Dominicus A. van Oeffelena
a Dow Benelux B.V., P.O. Box 48, NL-4530 AA Terneuzen, The Netherlands
b Centre for Catalytic Olefin Polymerisation, Stratingh Institute for Chemistry and Chemical Engineering,
University of Groningen, Nijenborgh 4, NL-9747 AG Groningen, The Netherlands
Received 29 April 2005; received in revised form 16 November 2005; accepted 22 December 2005
Available online 2 February 2006
Abstract
In the present study, the titanium-catalyzed ethylene trimerization in general, and more specifically, the concomitant PE formation have been
studied. The polymer formation is undesirable as it not only will lead to lower 1-hexene yields, but it can cause reactor fouling under the applied
conditions (30–80 ◦C, toluene solvent). It is, therefore, important to know which factors are involved in the formation of polymeric products and
how their formation can be reduced or even prevented. The PE formation turns out to be catalyzed by at least two different species. A significant
amount of PE is formed in the early stages of the reaction, caused by the presence of partly alkylated titanium species. The PE formation during
later stages of the reaction is due to degraded catalyst species, which means that polymer formation is an inherent property of this catalyst system.
The polymer output can be reduced largely by premixing CpꢀTiCl3 and methyl aluminoxane (MAO) prior to injection into the reactor. It was also
demonstrated that the type of MAO activator/impurity scavenger is of great importance. For a low yield of PE it is essential to use an MAO that
does not contain and/or is not able to generate aluminum hydride species. In the end the best results with respect to both productivity and selectivity
were obtained by starting from trimethyl titanium compounds.
© 2006 Elsevier B.V. All rights reserved.
Keywords: Catalysis; Ethylene; Polymerization; Titanium; Trimerization
1. Introduction
A commercially attractive route to 1-hexene is the selec-
tive trimerization of ethylene [1]. The only commercial process
capable of performing this conversion is based on a chromium
catalyst. ItisbeingpracticedinafacilityinQatar, whichisowned
by a joint venture of the Chevron Phillips Chemical Company
(CPChem) and Qatar Petroleum.
A titanium-based, highly active and highly selective cata-
lyst system for this conversion was discovered by Hessen and
coworkers [2]. The catalytic ethylene trimerization with the tita-
nium catalyst/methyl aluminoxane (MAO) system is a highly
efficient and very selective process, resulting in more then
90 wt% of C6 product from the converted ethylene, with an
excellent 1-hexene selectivity of more than 99%. The basic cat-
alyst system as published is [(␩5-C5H4CMe2Ph)TiCl3]/MAO
in toluene. Exploratory variations of the basic system indi-
cated that a CR2-bridge between the Cp and arene moieties
is essential for the trimerization selectivity, while substitu-
tions on the Cp-group can enhance the catalyst activity [2a].
Linear low-density polyethylene (LLDPE) is produced by the
copolymerization of ethylene with lower ␣-olefins, especially,
1-hexene and 1-octene. The market for LLDPE is predicted to
have an annual growth of 6% and as a consequence there is also
an increasing demand for lower ␣-olefins. Currently, the most
important process for the production of such ␣-olefins is the
catalytic oligomerizaton of ethylene to give a C4–C20 range of
linear ␣-olefins. Even though there is an increasing demand for
the lower ␣-olefins (C4–C8), producers are reluctant to invest in
new production units as they are faced with a much lower growth
for the higher ␣-olefins. Clearly, a need for more selective ␣-
olefin production processes exists.
∗
Corresponding author. Tel.: +34 977 559 013; fax: +34 977 559 379.
E-mail address: hhagen@dow.com (H. Hagen).
1381-1169/$ – see front matter © 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.molcata.2005.12.030

238
H. Hagen et al. / Journal of Molecular Catalysis A: Chemical 248 (2006) 237–247
Fig. 1. Catalysts structures and their numbering.
NMR spectra were recorded on a Varian Gemini 200 (1H:
200 MHz, 13C: 50.3 MHz) or a Varian VXR-300 (1H: 300 MHz,
The catalysts used in the present investigations are depicted in
Fig. 1.
1
13C: 75.4 MHz) spectrometer. The H and 13C NMR spectra,
This selectivity of the titanium-based system is higher than
for most of the other trimerization processes [3,4]. However, like
all other trimerization catalysts, the present one has a similar
disadvantage, i.e., the production of 2–5 wt% of high molecular
weight polyethylene. Not only will this lead to lower yields, but
the presence of high molecular weight PE in the reaction mixture
can cause reactor fouling. This is especially true for the titanium-
based process, which currently runs at lower temperatures than
the chromium-based systems (30–80 ◦C versus 110–125 ◦C for
the CPChem process [5]). It is, therefore, highly desirable to
know which factors are involved in the formation of polymeric
products and how the formation of these products can be reduced
or prevented. Here we report the results of our investigations.
measured at 25 ◦C, were referenced internally using the resid-
ual solvent resonances, and the chemical shifts (δ) are reported
in ppm. Gel permeation chromatography (GPC) analysis was
carried out on a Polymer Laboratories Ltd. (PL-GPC210) chro-
matograph at 135 ◦C using 1,2,4-trichlorobenzene as the mobile
phase. The samples were prepared by dissolving the poly-
mer (0.1% weight/volume) in the mobile phase solvent in an
external oven and were run without filtration. The molecular
weight was referenced to polyethylene (Mw = 50,000 g/mol) and
polystyrene (Mw = 100,000–500,000 g/mol) standards.
2.2. Electrospray ionization mass spectrometry
Electrospray ionization mass spectrometry (ES-MS) experi-
ments were conducted on a Nermag R3010 triple quadrupole MS
system with a custom-built IonSpray (pneumatically assisted
electrospray) source equipped with a gas curtain, comprised in a
closed chamber which can be evacuated, flushed and maintained
under nitrogen. Samples were taken up into a 500 ␮L syringe
(Model 1750 RNR, Hamilton) and electrosprayed via a syringe
pump operating at 10 ␮L/min. The capillary voltage was 3.5 kV.
Mass spectra were recorded from m/z 200 to 900 at 10 s per scan
under control of the Nermag Sidar data system. The sampling
orifice (nozzle) was increased from +40 to +160 V to generate
ion fragmentation. The skimmer located behind the sampling
orifice was at +25 V in all experiments. In a typical experiment,
2.7 mg (10 ␮mol) of 1a was dissolved in a mixture of 50 ␮L ␣-
olefin (1-hexene, 1-heptene or 1-octene) in 1 mL toluene. After
addition of 5.1 mg (10 ␮mol) B(C6F5)3, the sample was shaken
for 5 min and diluted 10 times with toluene generating a 10−3 M
solution. Details of the electrospray experiments can be found
in the supplementary information.
2. Experimentals
2.1. General considerations
All experiments were performed under a nitrogen atmo-
sphere using standard Schlenk techniques. Toluene, hexane
and pentane (Aldrich, anhydrous, 99.8%) were passed over
columns of Al2O3 (Fluka), BASF R3-11 supported Cu oxy-
gen scavenger and molecular sieves (Aldrich, 4 A). Diethyl
ether and THF (Aldrich, anhydrous, 99.8%) were dried over
Al2O3 (Fluka) and over molecular sieves (Aldrich, 4 A). All
solvents were degassed before use. Ethylene (AGA polymer
grade) was passed over BASF R3-11 supported Cu oxygen scav-
˚
˚
˚
enger and molecular sieves (Aldrich, 4 A). PMAO (4.9 wt% Al
in toluene, Akzo Nobel), PMAO-IP (13.4 wt% Al in toluene,
Akzo Nobel), MMAO-3A (2.03 wt% Al in isopar, Akzo Nobel),
TIBA (Witco), DIBAH (2.0 M in toluene, Aldrich) and AlMe3
(2.0 M in toluene, Aldrich) were used as received.
The compounds 6,6-pentamethylenefulvene [6], B(C6F5)3
[7] and catalysts 1, 3 and 5 [2b] were prepared according to
literature procedures. p-Tolyl- and 3,5-dimethylphenyl lithium
were synthesized according to standard procedures starting from
p-tolyl- or 3,5-dimethylphenylbromide by reaction with BuLi in
diethylether.
2.3. NMR tube experiments
All NMR tube experiments were prepared under a nitrogen
atmosphere in a glove box at room temperature and measured

H. Hagen et al. / Journal of Molecular Catalysis A: Chemical 248 (2006) 237–247
239
in Young-valve sealed NMR tubes. The mixtures were also ana-
lyzed by GC–MS. Details of the NMR tube experiments can be
found in the supplementary information.
ture was allowed to warm up to room temperature and stirred
for the next 12 h. The suspension was filtered, the residue
washed with neat diethyl ether and dried under reduced pres-
sure. After dissolving the pale yellow powder in 40 mL ice-
cooled ether/THF (5:1, v/v), 1.08 g (0.010 mol) trimethylsilyl
chloride was added. The white suspension was stirred at RT
for 2 h before it was cooled to 0 ◦C and 4.8 mL of 2.5 M
(0.012 mol) BuLi in hexane was added. After stirring for 5 h,
1.52 g (0.014 mol) trimethylsilyl chloride was added. The reac-
tion mixture was allowed to warm to room temperature and
stirred overnight. The volatiles were removed and the residue
was three times extracted with 15 mL CH2Cl2. The combined
extracts were filtered through silica before the solvent was
removed, leaving the 3-[1-(3,5-dimethylphenyl)cyclohexyl]-
bistrimethylsilylcyclopentadiene as orange oil, which was used
without further purification.
The oil was dissolved in 30 mL neat CH2Cl2 and cooled to
−20 ◦C before 1.5 g (0.008 mol) TiCl4 was added. The mixture
was allowed to warm up to room temperature and stirred for
12 h. After removal of all volatiles the residue was continuously
extracted with 30 mL of hexane. Cooling of the hexane extract to
−30 ◦C gave 1.10 g of pure 4 as red crystals in 23% overall yield.
1H-NMR (δ, C6D6): 0.10 (s, 9H), 1.20–1.50 (m, 6H),
1.78–2.10 (m, 2H), 2.10 (s, 6H), 2.64 (m, 2H), 6.41 (t, 1H,
3JHH = 3.1 Hz), 6.47 (t, 1H, 3JHH = 3.1 Hz), 6.64 (br, 1H), 6.96
(br, 1H), 7.12 (br, 2H).
2.4. General description of trimerization experiments
The catalytic ethylene trimerization reactions were per-
formed in a stainless steel 1 L autoclave (Medimex) in batch
or semi-batch mode. The reactor was temperature and pressure
controlled and in case of semi-batch experiments the pres-
sure was kept constant to within 0.2 bar of the initial pressure
by addition of ethylene. After the desired reaction time the
reactor was vented and residual MAO was destroyed by addi-
tion of 20 mL of ethanol. Polymeric product was collected,
stirred for 90 min in acidified ethanol and rinsed with ethanol
and acetone on a glass frit. The polymer was initially dried
on air and subsequently in vacuum at 80 ◦C. Details of the
trimerization experiments can be found in the supplementary
information.
2.5. Synthesis of [1-(4-methylphenyl)cyclohexyl]-
cyclopentadienyl titanium trichloride (2)
0.98 g (0.010 mol) p-tolyl lithium was dissolved in 50 mL
of ice-cooled diethyl ether before 1.46 g (0.010 mol) 6,6-
pentamethylenefulvene was added. The mixture was allowed to
warm up to room temperature and stirred for 12 h. The suspen-
sion was filtered, the residue washed with neat diethyl ether and
dried under reduced pressure. After dissolving the pale yellow
powderin20 mLice-cooledTHF, trimethylsilylchloride(1.20 g,
0.011 mol) was added. The mixture was stirred at room temper-
ature for 2 h before all volatiles were removed under reduced
pressure. The residue was suspended in 20 mL CH2Cl2 and
filtered through silica. Removal of the solvent leaves the 3-[1-(4-
methylphenyl)cyclohexyl]trimethylsilylcyclopentadiene as yel-
low oil, which was used without further purification.
2.7. Synthesis of 1-phenylcyclohex-1-yl-cyclopentadienyl
trimethyl titanium (1a)
0.4 g (1.9 mmol) of 1-phenylcyclohex-1-yl-cyclopentadienyl
titanium trichloride (1) was dissolved in 20 mL ice-cooled THF
and 1.2 mL of a 3.0 M solution of MeMgCl in THF was added.
After stirring the suspension for 4 h, all volatiles were removed
under reduced pressure. The residue was extracted with 20 mL
of pentane. The pentane solution was filtered and the solvent was
pumped off to give a colorless solid which was re-crystallized
from 5 mL neat pentane to give 250 mg white crystals of 1-
phenylcyclohex-1-yl-cyclopentadienyl trimethyl titanium (1a)
in 73% yield.
The yellow oil was dissolved in 30 mL neat CH2Cl2 and
cooled to −20 ◦C before 1.5 g (0.008 mol) TiCl4 was added.
The mixture was allowed to warm up to room temperature and
stirred for 12 h. After removal of all volatiles the residue was
continuously extracted with 30 mL of hexane. Cooling of the
hexane extract to −30 ◦C gave 1.60 g of pure 2 as bright orange
crystals in 38% overall yield.
1H-NMR (δ, C6D6): 1.10–1.40 (m, 6H), 1.23 (s, 9H), 1.64 (dt,
2H, 3JHH = 12.5 Hz, 3JHH = 3.3 Hz), 2.40 (3JHH = 12.5 Hz), 5.75
3
3
(t, 2H, JHH = 2.8 Hz), 5.84 (t, 2H, JHH = 2.8 Hz), 7.10–7.26
3-[(4-MeC6H4)C6H10]C5H4SiMe3: 1H-NMR (δ, CDCl3):
−0.06 (s, 9H), 1.40–1.60 (m, 6H), 2.10 (m, 4H), 2.29 (s,
3H), 3.22 (s, 1H), 6.10 (s, 1H), 6.38 (m, 2H), 7.07 (d, 2H,
3JHH = 8.0 Hz), 7.22 (d, 2H, 3JHH = 8.0 Hz).
(m, 5H).
2.8. Synthesis of 3-[2-(3,5-dimethylphenyl)prop-2-yl]-
trimethylsilylcyclopentadienyl trimethyl titanium (5a)
2: 1H-NMR (δ, CDCl3): 1.20–1.50 (m, 6H), 1.89 (t, br, 2H),
2.12 (s, 3H), 2.52 (d, br, 2H), 5.94 (t, br, 2H), 6.34 (t, br, 2H),
6.67 (d, 2H, 3JHH = 8.0 Hz), 7.22 (d, 2H, 3JHH = 8.0 Hz).
3-[2-(3,5-Dimethylphenyl)prop-2-yl]-trimethylsilylcyclo-
pentadienyl titanium trichloride (5) (0.3 g, 0.69 mmol) was
dissolved in ice-cooled diethyl ether/THF (20 mL, 50/50 vol%)
and MeMgCl (0.69 mL, 3.0 M in THF) was added. After stirring
the suspension for 4 h all volatiles were removed under reduced
pressure. The residue was extracted with 20 mL of pentane.
The pentane solution was filtered and the solvent was pumped
off to give a colorless oil of 3-[2-(3,5-dimethylphenyl)prop-2-
yl]-trimethylsilylcyclopentadienyl trimethyl titanium (5a) in
2.6. Synthesis of 3-[1-(3,5-dimethylphenyl)cyclohexyl]-
trimethylsilylcyclopentadienyl titanium trichloride (4)
1.12 g (0.010 mol) 3,5-dimethylphenyl lithium was dis-
solved in 50 mL of ice-cooled diethyl ether before 1.46 g
(0.010 mol) 6,6-pentamethylenefulvene was added. The mix-

240
H. Hagen et al. / Journal of Molecular Catalysis A: Chemical 248 (2006) 237–247
(b) GC–MS (min[M•+]): 7.36 [120, ENB], 7.42 [120, ENB],
8.30 [136], 13.90 [224, C5H5C(Ph)C5H10], 14.40 [256].
(c) GC–MS (min[M•+]): 7.33 [120, ENB], 7.40 [120, ENB],
8.2.5 [136], 13.42 [248], 13.53 [248], 13.70 [246], 13.90
[224, C5H5C(Ph)C5H10], 14.40 [256], 14.46 [256], 17.08
[352].
almost quantitative yield. After standing for a few weeks at
0 ◦C the oil became crystalline.
1H-NMR (δ, C6D6): 0.11 (s, 9H), 1.34 (s, 9H), 1.48 (s, 3H),
1.51 (s, 3H), 2.16 (s, 6H), 6.12 (m, 1H), 6.19 (m, 1H), 6.24 (m,
1H), 6.69 (s, 1H), 6.93 (s, 1H).
(d) GC–MS (min[M•+]): 7.33 [120, ENB], 7.40 [120, ENB],
8.2.5 [136], 12.86 [234], 12.97 [234], 13.12 [232], 13.22
[232], 13.90 [224, C5H5C(Ph)C5H10], 14.40 [256], 14.46
[256], 14.53 [256], 16.61 [338], 16.63 [338].
2.9. Premix experiments
2.9.1. Reaction of 1 with MAO
19 mg (50 ␮mol) of 1 was added to 2.8 g (500 ␮mol) MAO
(4.9 wt% Al in toluene) in 2 mL of toluene. The sample was
shaken and allowed to settle. After a few minutes a blue solid
was formed, which turned into a blue oil after an additional
5 min. The MAO/toluene solution was decanted of before 2 mL
neat toluene and 0.5 mL methanol were added. After filtration
over silica the sample was analyzed by GC–MS.
3. Results and discussion
The trimerization catalysis is thought to start with a tita-
nium(IV) catalyst precursor that upon activation with methyl
aluminoxane forms a cationic dimethyl complex. In this com-
plex, the arene moiety of the ligand coordinates to the metal
center. The dimethyl species can undergo ethylene insertions
to produce a bis(n-alkyl) compound. Supposedly, this type of
complex is in equilibrium with alkyl–olefin–hydride species,
which would have the arene moiety detached. Displacement of
the olefin by ethylene, which rapidly inserts into the M H bond,
would be identical to the normal chain transfer process of cat-
alytic ethylene polymerization. However, the olefin can also be
displaced by the pendant arene moiety to yield an alkyl–hydride
complex, which then can undergo reductive elimination to give
an alkene and a Ti2+ species. The latter reaction has been pro-
posed as the key step for the switch from a polymerization
to a trimerization catalyst. Coordination and coupling of two
molecules of ethylene gives a titana(IV) cyclopentane. Migra-
tory insertion of a third molecule of ethylene results in the
formation of a titana(IV) cycloheptane. Subsequent reductive
elimination and displacement steps similar to those occurring in
the dialkyl species as discussed above, give back the titanium(II)
intermediate (see Scheme 1) [2a,2b,8].
GC–MS (min[M•+]): 11.01 (160, cyclohexyl-benzene),
11.55 (158, 1-phenyl-cyclohexene), 13.95 [224, C5H5C(Ph)
C5H10].
2.9.2. Reactions of 1 with MAO in presence of an olefin
3.8 mg (10 ␮mol) of 1 was added to a mixture of 1.375 g
(250 ␮mol) MAO (4.9 wt% Al in toluene) and 2 mL of toluene
containing (a) 0.05 mL 1-hexene, (b) 0.05 mL ENB, (c) 0.05 mL
ENB and 0.05 mL 1-heptene or (d) 0.05 mL ENB and 0.05 mL
1-octene. The samples were shaken for 5 min before 0.5 mL of
methanol was added. After filtration over silica, the sample was
analyzed by GC–MS.
(a) GC–MS (min[M•+]): 3.17 [98, 2-methyl-1-hexene], 3.40
[98, Z-3-heptene], 3.48 [98, 2-methyl-2-hexene], 9.42
[166], 9.86 [168], 9.99 [168, Z-2-dodecene], 10.03 [168],
10.33 [168], 11.51 (158, 1-phenyl-cyclohexene), 12.57
[252], 12.77 [252], 13.30 [250], 13.40 [252], 13.90 [224,
C5H5C(Ph)C5H10].
Scheme 1. Proposed reaction mechanism for the catalyst activation and ethylene trimerization.

H. Hagen et al. / Journal of Molecular Catalysis A: Chemical 248 (2006) 237–247
241
methanewasobserved, whilein theresiduepredominantlyeither
2-methyl-1-hexene together with traces of methyl-dodecene or
2-methyl-1-octenewasdetected. Apparently, thepresenceofone
alkyl with ␤-hydrogens and one methyl group at the cationic
titanium(IV) center results in a fast ␤-hydrogen transfer, which
should lead to the generation of the active trimerization catalyst.
Even though it cannot be excluded that during the activation at
high ethylene pressures PE is formed during the activation step,
it is very likely that the majority of the initial polymer originates
from other species.
The original reaction mechanism (Scheme 1) stated that the
titanium trichloride precatalysts react with MAO to form the
corresponding dimethyl titanium cations [2a,2b]. It has always
been assumed that this reaction is fast and quantitative. In
order to investigate if this is actually the case, the alkylation
of cyclopentadienyl titanium trichlorides with aluminum alkyls
was studied by means of NMR tube experiments. Reacting
CpꢀTiCl3 with 2–5 eq. of AlMe3 in deuterobenzene yielded only
the monomethyl complex CpꢀTiMeCl2. This result is in agree-
ment with the earlier reported synthesis of Cp*TiMeCl2, which
was done by using AlMe3 as a selective mono-alkylating agent
[9]. The reaction with 10–30 eq. of AlMe3 gave the dimethyl
derivative CpꢀTiMe2Cl. Both the mono- and dimethyl com-
pounds CpꢀTiMeCl2 and CpꢀTiMe2Cl were found to be stable
for at least 24 h at room temperature in aromatic solvents con-
taining AlMe3. Similar observations were made for the reactions
of CpꢀTiCl3 with 2–30 eq. of MAO. In all cases, the presence of
trimethyl titanium complexes could not be detected.
Fig. 2. 1-Hexene and PE formation as a function of time for 4/MAO (3 ␮mol
catalyst, 260 mL toluene, 15 bar ethylene, Ti:Al = 1:500 (m/m), 30 ◦C). Lines
are intended for illustrative purposes only.
During the trimerization experiments always the formation of
about 2–5 wt% of linear high-density polyethylene was observed
[2a]. To gain more insight into the reason of the polyethylene
formation, the 1-hexene and PE formation profiles were deter-
mined for catalyst 4/MAO (see Fig. 2). At 30 ◦C and 15 bar of
ethylene pressure this catalyst does not show any significant
deactivation over a 3-h period. As shown in Fig. 2, it is possible
to identify three phases in the polymer formation process, i.e.,
(1) a short initial activation period, during which the polymer
formation rate is high, (2) a long period, with a moderate growth
of polymeric product and (3) a final period with an increased rate
in polymerization activity. From the shape of the profile for this
stable trimerization catalyst we can conclude, that there are at
least two, but perhaps three catalyst species that give polymeric
products. This can also be concluded from the GPC analyses
of the polyethylene formed during the trimerization reactions.
Typically, Mw is in the range of 500,000–1,000,000 g/mol with
molecular weight distributions (MWD) of around 40. For a
single-site polymerization catalyst under constant conditions
(temperature and ethylene concentration), the MWD would have
Flory’s theoretical value of 2. The fact that the MWDs are much
broader proofs that multiple species are active.
Reacting CpꢀTiCl3 with 50 eq. of AlMe3 in C6D6 resulted
in the immediate formation of the dimethyl complex, which
slowly reacted further to form a violet-colored solid with the
concomitant release of methane. The nature of violet-colored
solid is unclear, but based on the color it is presumably a tita-
nium(III) species. The intermediate compound, from which the
titanium(III) species is formed, is probably the trimethyl tita-
nium(IV) complex as an NMR tube experiment demonstrated
that in absence of an olefin, but in the presence of AlMe3 the
trimethyl titanium complex CpꢀTiMe3 (1a) is not stable and
decomposes while releasing a gas (most likely methane).
Theconventionalprocedureforstartingatrimerizationexper-
iments is injection of the trichloro titanium species, CpꢀTiCl3,
into the reactor loaded with solvent, ethylene and MAO. The
slow reaction of CpꢀTiMe2Cl to CpꢀTiMe3 makes it likely that
the partly alkylated compound is present in the presence of both
ethylene and excess of MAO. Methyl or chloride abstraction
from this compound will then generate either the dimethyl tita-
nium cation or the cationic methyl titaniumchloride species.
Whereas the former, according to the proposed reaction mecha-
nism (see Scheme 1), will react to the titanium(II) trimerization
species, the latter species is an active ethylene polymerization
catalyst [9,10]. When present such a species will result in the
formation of polymer during in the early stages of the reaction.
3.1. Polymer formation in the activation period
As can be seen in Fig. 2 a significant part of the polymer is
formed in an early stage of the reaction. Originally, we assumed
that a source of PE is the intermediate dialkyl titanium(IV)
cationic species (see Scheme 1). Prior to the changeover to the
titanium(II) trimerization catalyst, such a species could catalyze
the polymerization of ethylene to PE. However, when consider-
ing the results of the DFT calculations by Blok and Budzelaar
[8a], one could question the validity of this mechanism. These
authorsshowedthatforadiethyltitaniumcationthedirecthydro-
gen transfer is already favored over the next ethylene insertion,
which means that polymer formation is unlikely. The outcome
of these calculations was supported by NMR tube experiments.
After adding B(C6F5)3 to a mixture of a trimethyl cyclopen-
tadienyl titanium species and a few equivalents of 1-hexene
or 1-octene in an aromatic solvent, spontaneous evolution of
3.2. Polymer formation in the second and third period
As mentioned before, after the rapid polymer formation at
the beginning of a run, a slow but continuous polymer for-

242
H. Hagen et al. / Journal of Molecular Catalysis A: Chemical 248 (2006) 237–247
Table 1
increase in the quantity of polyethylene. PMAO and d-MAO
differ only in the amount of free AlMe3 and based on the simi-
larity in results, it is justified to conclude that the trimerization
behavior is relatively independent on the amount of free AlMe3
[13]. What makes these two activators different from the oth-
ers is the absence of aluminium hydrides. It appears that the
use of MAOs, which either contain aluminum hydride species
(MMAO-3A, PMAO-IP, PMAO + DIBAH) or are able to gener-
ate them (PMAO + TIBA), affects both productivity and selec-
tivity. This behavior might be the consequence of a stronger
coordinating ability of the hydride containing anions versus
the methyl containing ones [14]. As a stronger coordination
means more blocking of the active sites a lower productivity
will be observed. In this context it is notable that both Arndt et
al. [14] and Muhoro and Hartwig [15] were able to get X-ray
structure determinations of a bis(cyclopentadienyl)titanium(II)
complexes with a coordinated hydroborane or diisobutyl alu-
minum hydride, respectively.
The decrease in selectivity probably has its origin in the
alkylating power of the various MAOs. As demonstrated above
partially alkylated titanium(IV) species, such as CpꢀTiMe2Cl,
are thought to be one of the main reasons for polymer forma-
tion (at least in the earlier stages of the reaction). If a certain
type of MAO is less active in the alkylating reaction, this will
lead to an increase in PE formation. In addition, it will also lead
to a lower production of 1-hexene as the absolute number of
species active in trimerization is less. The difference in activ-
ity in the alkylation reaction can also explain why PMAO-IP
performs so much worse than PMAO. With respect to its com-
position it differs from PMAO that it has less free AlMe3, and
that it contains some hydrides. Based on the results for d-MAO,
the lower free AlMe3 content should not make a substantial dif-
ference. Also, the hydride content is quite low (about 15% of
MMAO-3A), which makes the reasoning that hydride coordi-
nation alone is causing its low productivity unsatisfactory. A
very distinct difference between PMAO and PMAO-IP resides
in the synthetic procedure. Contrary to PMAO, PMAO-IP is
prepared via a non-hydrolytic route [16]. As a consequence
of this different synthetic route, PMAO-IP has an other cage-
structure resulting in a higher storage stability. Unlike PMAO it
does not form a precipitate with time, but it gives solutions with
increased viscosities. The precipitate in the case of PMAO is an
aluminium oxide-like structure, resulting from AlMe3 liberation
from the cages. It is quite possible that this difference in stabil-
ity is translated into a difference in reactivity in the alkylation
reaction.
Influence of co-catalyst type on productivity and selectivitya
Co-catalyst
C6 (g)
PE (g)
Productivity (103 kg
C6/PE ratio
(g/g)
C6 mol−1 Ti h−1
)
PMAO
d-MAO
PMAO + TIBA
PMAO + DIBAH
PMAO-IP
54.0
52.9
35.9
31.1
33.9
35.7
1.3
1.2
2.1
2.2
3.4
1.6
11.4
11.1
7.6
6.5
7.1
42
44
17
14
10
22
MMAO-3A
7.5
a
Batch mode, 3 ␮mol catalyst 5, Ti:Al = 1:500 (m/m), 360 mL toluene, 50 ◦C,
30 bar ethylene, 95 min run time.
mation is observed. Such slow processes are not easily inter-
preted. For example it is not possible to distinguish between
very small amounts of highly active short-living catalysts or
slow but continuously active ones. Nevertheless, several cata-
lyst transformation pathways can be envisaged with as the most
important one the degradation of the ligand system. Evidence
for this route came from a number of experiments with pre-
mixed catalyst–olefin systems (vide infra). After hydrolysis of
such a mixture containing catalyst 1, MAO and 1-hexene not
only the expected 2-methylhexene, and free ligand were found,
but in addition 1-phenylcyclohexene was detected as well. The
latter is a fragmentation product of the substituted cyclopenta-
dienyl ligand. Loss of the aryl-bearing substituent (instrumental
in achieving good trimerization selectivity) will leave an unsub-
stituted cyclopentadienyl titanium species, which is likely to be
active as an ethylene polymerization catalyst.
To evaluate the influence of the type of MAO activator
on polymer formation trimerization reactions were carried out
in batch mode with different kinds of commercially available
MAOs (Akzo Nobel [11]) and with mixtures of MAO and alkyl
aluminum compounds:
• PMAO in toluene (Akzo Nobel [11])
◦ 4.9 wt% Al, 25 wt% free AlMe3.
◦ Composition of hydrolysis gas: 98.7 mol% CH4, 0.0 mol%
H2.
• Vacuum-dried PMAO (d-MAO) [12].
• PMAO (450 eq.) + triisobutyl aluminum (TIBA) (50 eq.).
• PMAO (450 eq.) + diisobutyl aluminum hydride (DIBAH)
(50 eq.).
• PMAO-IP in toluene (Akzo Nobel [11])
◦ 13.4 wt% Al, 14.0 wt% free AlMe3.
◦ Composition of hydrolysis gas: 99.6 mol% CH4, 0.2 mol%
H2.
• MMAO-3A in isopar (Akzo Nobel [11])
◦ 2.1 wt% Al, ∼40 wt% free AlR3.
◦ Composition of hydrolysis gas: ∼70 mol% CH4, 30 mol%
isobutane, 1.5 mol% H2.
3.3. Premixing
As discussed above partially alkylated titanium species are
assumed to be responsible for the formation of polymer. By
using catalysts that have been alkylated prior to their addition
to the reactor, it should be possible to avoid the polymer forma-
tion by these species. The best way of doing this, is, of course,
the synthesis of trimethyl titanium species, CpꢀTiMe3. However,
the synthesis and isolation of these compounds is not always
straightforward. Another approach would be the reaction of
As presented in Table 1, the C6 productivity and C6/PE ratio
were found to be highly sensitive for the type of MAO acti-
vator used. Whereas PMAO and d-MAO performed equally
well with respect to productivity and selectivity, the use of
the other MAOs resulted in both a lower productivity and an

H. Hagen et al. / Journal of Molecular Catalysis A: Chemical 248 (2006) 237–247
243
Table 2
the catalyst with an unsubstituted aryl group (1 and 3) a >70%
increase in productivity was observed.
The decrease in productivity by using longer premix times
(15 min instead of 1 min) clearly shows that the titanium alkyls
are not very stable under the conditions used and that the premix
time should be short.
Effect of premixing step on productivity and selectivitya
Catalyst Methodb C6 (g) C10 (g) PE (g) Productivity
C6/PE
ratio
(g/g)
(103 kg C6
mol−1 Ti h−1
)
Standard 13.4
0.5
0.8
1.1
0.7
5.3
7.1
13
25
Premix
17.6
1
Instead of using MAO as the alkylating agent in the premix
step, it is also possible to use AlMe3. By injecting catalyst solu-
tions, which were prepared by pre-mixing the cyclopentadienyl
titanium trichlorides with 50 eq. of AlMe3, into toluene solu-
tions containing MAO (250 or 450 eq.), a reduction of the PE
formation was observed in comparison to the standard proce-
dure (cf. catalyst 1 – premix, Table 2 with entry 1, Table 3).
Under inverted conditions, i.e., a catalyst premixed with 250 eq.
of MAO and 50 eq. of TMA as a scavenger (entry 2), very similar
results were observed. In the latter case a lower 1-hexene pro-
ductivity was observed than for the MAO-only case. The reason
for this behavior is most likely the lower amount of impurity
scavenger present during these runs in comparison to the base-
line run. Similar effects can be observed by looking at results
for a standard run with less MAO (300 eq. versus 500 eq., entry
3, lower productivity) or a premix run with more MAO in the
reactor (450 eq. versus 250 eq., entry 4, higher productivity).
Also when using a different catalyst, 3, increasing the amount
of MAO improves the efficiency while the amount of PE formed
remains more or less the same (entries 5 and 6).
The importance of the alkylation level was further proven by
trimerization experimentsthat werestarted withtitaniumspecies
having different degrees of alkylation (see Table 4). By using the
trimethyl versions of catalysts 1 and 5 (1a and 5a), the catalytic
productivity was almost doubled, while the selectivity was dou-
bled (5) or even almost quadrupled (1).
From the aforementioned results, it is clear that a large part
of the polyethylene formation can be suppressed by activating
the pre-catalyst in the absence of ethylene. The premix time
should be short as prolonged exposure of the catalyst to the alky-
lating agent without ethylene present reduces the productivity,
even though the polymer formation itself is reduced as well and
the C6/PE ratio is improved. The extent of the decrease in pro-
ductivity is much larger than the one observed normally during
trimerization experiments. This suggests that the species active
in ethylene trimerization is not stable in the absence of ethylene.
One way to achieve the advantage of the premixing without this
loss of productivity could be premixing of pre-catalyst and MAO
Premixb
9.1
0.2
0.3
3.6
35
Standard
Premix
2.0
3.0
0.0
0.0
0.4
0.3
1.2
1.2
8
11
2
3
4
5
Standard
Premix
9.1
19.5
0.1
0.5
0.4
0.3
12.1
26.0
23
65
Standard 10.2
Premix 11.8
0.1
0.1
0.2
0.1
13.7
15.8
57
84
Standard 15.3
Premix 19.2
0.5
0.8
0.5
0.5
20.4
25.5
31
38
a
Semi-batch mode, 260 mL toluene solvent, 30 ◦C, 10 bar ethylene, 15 min
run time, 10 ␮mol for catalysts 1 and 2, 3 ␮mol for catalysts 3, 4 and 5, Ti:total
Al = 1:500 (m/m).
b
15 min premix time.
trichloro titanium species, CpꢀTiCl3, with MAO outside the reac-
tor, and this procedure has been used to test the hypothesis. The
results have been listed in Table 2. To eliminate other variables
than pre-alkylation (e.g., decomposition, ethylene/1-hexene co-
trimerization, temperature) on the catalyst performance a short
run time of 15 min and a low ethylene pressure (10 bar) were
chosen.
From the results in Table 2, it is clear that the pre-activation
of the catalysts in all cases had a beneficial influence on the 1-
hexene/PE ratio. An increase in the 1-hexene productivity was
observed, while the PE formation was somewhat reduced or at
the most equal. Both results indicate that by premixing more
trimerization catalyst species are formed in the activation step
at the expense of the number of polymerization catalyst species.
The increase in productivity upon premixing with MAO is
strongly related to the catalyst structure. Catalysts with substi-
tuted pendant aryl groups showed only a low (2) or a moderate
(4 and 5) increase in productivity after pre-mixing, while for
Table 3
Effect of type and amount of co-catalyst and method of addition on productivity and selectivitya
Entry
Premix (eq.)
Reactor (eq.)
C6 (g)
C10 (g)
PE (g)
Productivityc
C6/PE ratio (g/g)
1
2
3
4
5b
6b
AlMe3 (50)
MAO (250)
–
AlMe3 (50)
MAO (250)
MAO (750)
MAO (250)
AlMe3 (50)
MAO (300)
MAO (450)
MAO (250)
MAO (750)
13.4
12.9
10.0
15.5
19.5
27.8
0.4
0.3
0.3
0.7
0.5
1.2
0.6
0.6
0.7
0.6
0.3
0.4
5.4
5.2
4.0
6.2
26.0
37.1
22
22
13
26
65
70
a
Semi-batch mode, catalyst 1 (10 ␮mol), 260 mL toluene solvent, 30 ◦C, 10 bar ethylene, 15 min run time, 1 min premix time.
Catalyst 3 (3 ␮mol).
b
c
Productivity in 103 kg C6 mol−1 Ti h−1
.

244
H. Hagen et al. / Journal of Molecular Catalysis A: Chemical 248 (2006) 237–247
Table 4
Effect of the alkylation stage of added catalyst on productivity and selectivitya
Catalyst
C6 (g)
C10 (g)
PE (g)
Productivity (103 kg C6 mol−1 Ti h−1
)
C6/PE ratio (g/g)
CpꢀTiCl3 (1)
13.1
15.5
23.1
15.3
31.9
0.5
0.7
0.9
0.5
2.0
1.0
0.6
0.5
0.5
0.5
5.2
6.2
9.2
20.4
42.3
13
26
46
31
60
CpꢀTiMe2Clb
CpꢀTiMe3 (1a)
CpꢀꢀTiCl3 (5)
CpꢀꢀTiMe3 (5a)c
a
Semi-batch mode, 260 mL toluene solvent, 30 ◦C, 10 bar ethylene, 15 min run time, catalyst amount: 10 ␮mol (1), 3 ␮mol (5), Ti:Al = 1:500 (m/m).
CpꢀTiCl3 (1) was pre-mixed with 50 eq. of AlMe3 for 1 min, total Ti:Al = 1:500 (m/m).
Activated with 1.1 eq. of [R3N][B(C6F5)3], MAO (Ti:Al = 1:20 (m/m) present as impurity scavenger.
b
c
in the presence of alkenes other than ethylene. Such alkenes, e.g.,
expected methane and 2-methylhexene, also the hexene isomers
2- and 3-hexene were observed. This shows that in the absence
of ethylene the titanium trimerization catalysts are olefin iso-
merization catalysts. The isomerization of 1- to 2-hexene (or
identically from 2- to 3-hexene) will most likely proceed via a
mechanism involving titanium allyl hydride species (Scheme 2,
B) and oxidative addition/reductive elimination reactions. The
1-hexene side-on coordinated species (Scheme 2, A) is identical
to the complex formed in the last step of the catalytic circle of
the trimerization reaction (see Scheme 1). If the life-time of this
complex is sufficiently long, i.e., if the exchange with ethylene
is not immediate, ␥-hydrogen abstraction of the 1-hexene results
in the formation of an allyl-hydride complex with the potential
of forming a titanium(III) hydride species (D), which is known
as an active polymerization catalyst [17]. This mechanism was
supported by the results from electrospray ionization mass spec-
troscopy (ESI-MS) analyses [18]. For our investigations solu-
tions of 1a in aromatic solvents were reacted with 1.1 eq. of
B(C6F5)3 or 20 eq. of MAO in the presence of 10 eq. of an ␣-
1-hexene, would have the same stabilizing effect on the active
catalyst, but even if polymerization would occur during the acti-
vation, asolublepolymer(e.g., poly-1-hexene)wouldbeformed.
In order to check the hypothesis a number of olefins were tested
by premixing a catalyst with 250 eq. of MAO activator in the
presence of an olefin (a complete overview of all results is given
in the supplementary information). In all cases, norbornene and
its derivatives enhance productivity, while a good selectivity is
maintained. For other olefins there is no systematic change, as
in some cases the presence of olefins has hardly any influence,
while in others cases a strong decrease in both productivity and
selectivity was observed. This striking difference between 2-
norbornene or one of its analogs and the other olefins is very
interesting and gives possible clues about the underlying rea-
sons for polymer formation during the trimerization process.
To investigate the reactions taking place during the preactiva-
tion in the presence of an olefin, a mixture containing 1-hexene
was hydrolyzed and analyzed with GC–MS. In addition to the
Scheme 2. Possible reactions for 1-hexene coordinated Ti-species.

H. Hagen et al. / Journal of Molecular Catalysis A: Chemical 248 (2006) 237–247
245
Fig. 3. Saturated and unsaturated hydrolysis products of preactivation mixtures
containing ENB.
Fig. 4. Co-trimerization products for norbornene derivatives.
olefin. By using a mixture of 1-hexene, 1-heptene or 1-octene as
olefins, mass peaks corresponding to the [(C5H4CMe2Ph)TiH]+
+
and {[(C5H4CMe2Ph)Ti(1-alkene)]–H} cations (Scheme 2, D
2,5-norbornadiene, a rather long (0.05 mL: 10–15 s, 0.5 mL:
3–4 min) initiation period (the time between catalyst injection
andobservedreactorexotherm)ispresent. Thisphenomenoncan
be explained by the assumption that the ethylene trimerization
to 1-hexene does not start until all of the norbornene derivatives
have been converted into the corresponding co-trimer. The pres-
ence of an olefin, which binds stronger to the titanium metal
center than ethylene, will induce a selective co-trimerization
that is preferred over ethylene trimerization. Interestingly, this
means that the titanium ethylene trimerization catalyst may
be useful for the preparation of some unique ␣-olefins, such
as 4-norbornyl-1-butene and related compounds (see Fig. 4)
[20]. A comparable selective co-trimerization process of ethy-
lene and styrene was also observed by Pellechia et al. [21] for
the (C5Me5)TiMe3/B(C6F5)3 catalyst in toluene during their
attempts to copolymerize these olefins.
From the present study it has become clear that the polymer
formation can be reduced drastically to less then 1 wt%. Unfor-
tunately, the remaining small amount of high-molecular weight
polymer could still lead to issues, e.g., reactor fouling. One way
to reduce the disadvantageous effects of the polymeric prod-
ucts is, instead of preventing the polymerization itself, lowering
of the molecular weight of the polyethylene formed. From the
ethylene trimerization reactions catalyzed by chromium-based
systems it is known that addition of hydrogen can reduce the
molecular weight of the polymer to such an extent, that it will
stay in solution under the applied temperature of about 100 ◦C
[3d,4,22]. In order to evaluate the influence of hydrogen on the
present system, the trimerization reaction was carried out for
two catalysts under a partial hydrogen pressure of 3 bar (see
Table 5). However, no positive effect on the polymer forma-
tion/solubility was observed. The molecular weight of the PE
remained the same, pointing to a very low H2 response of the
catalysts. Obviously, the productivity decreased as the ethylene
partial pressure, and thus the ethylene concentration in solution,
was lower.
and B) were observed [19]. An important conclusion from this
is, is that, fundamentally, it will be impossible to reduce the
polymer formation completely.
The reason that the preactivation in the presence of nor-
bornene or one of its derivatives never leads to poorer results is
caused by the difference in the activation mechanism. Similarly
to the other olefins, 2-norbornene and its analogs can insert in the
Ti Me bond. The second step, the direct hydrogen transfer, is
however disfavored due to steric reasons and the titanium com-
plex will remain in the 4+ oxidation state. This was confirmed
by GC–MS analysis of hydrolyzed mixtures of 1/MAO/ENB,
which showed only compounds with saturated end-groups, such
as 2-methyl-5-ethylidene-norbornane (E, Fig. 3) or 2-methyl-3-
(5-ethylidene-norborn-2-yl)-5-ethylidene-norbornane (F). The
formation of titanium(II) species will only occur after an inser-
tion of an ethylene (or an ␣-olefin) as was demonstrated
by similar GC–MS analyzes of mixtures of 1/MAO/ENB
and an ␣-olefin (1-heptene, 1-octene). In addition to com-
pounds E and F the compounds 2-methyl-3-(hepten-2-yl)-
5-ethylidene-norbornane (G) and 2-methyl-3-(octen-2-yl)-5-
ethylidene-norbornane (H) were found. The latter, which both
have an additional unsaturation, derive from species that have
undergone an initial insertion of ENB into the titanium methyl
bond, followed by an insertion of the ␣-olefin and subsequent
hydrogen transfer. This confirms that the Ti catalyst activated
in the presence of 2-norbornene or ENB remains in the 4+ oxi-
dation state, while in case of other olefins after the insertion
of the olefin it reacts further to titanium(II). After an insertion
of ethylene, which in the case of preactivation will happen in
the reactor, the titanium(IV) precatalyst will be transformed to
the actual titanium(II) catalyst species. Therefore, there is no
possibility to rearrange to a titanium(III) compound with the
accompanying increase in PE formation.
It was noted that in the trimerization experiments with
catalysts premixed in the presence of ENB, 2-norbornene or

246
H. Hagen et al. / Journal of Molecular Catalysis A: Chemical 248 (2006) 237–247
Table 5
Hydrogen influence on the titanium catalyst trimerizationa
Catalyst (␮mol)
H2 (bar)b
C6 (g)
C10 (g)
PE (g)
Productivity (103 kg C6 mol−1 Ti h−1
)
1
(15)
–
3
55.1
44.7
4.1
3.3
4.6
3.8
2.3
1.9
5
(3)
–
3
98.7
82.9
11.9
8.4
1.9
2.4
20.8
17.5
a
Semi-batch mode, 30 ◦C, 260 mL toluene, 15 bar total pressure, 95 min run time, Ti:Al = 1:500 (m/m).
Partial pressure.
b
4. Conclusions
[2] (a) P.J.W. Deckers, B. Hessen, J.H. Teuben, Angew. Chem. Int. Ed. 40
(2001) 2516;
(b) P.J.W. Deckers, B. Hessen, J.H. Teuben, Organometallics 21 (2002)
5122;
(c) B. Hessen, P.J.W. Deckers, WO 2002066405 A1 (2002) to Stichting
Dutch Polymer Institute.
The catalytic ethylene trimerization with the titanium cata-
lyst/MAO system is a highly efficient and very selective process,
resulting in more then 90 wt% of C6 product from the converted
ethylene, with an excellent C6 selectivity of more than 99%
towards 1-hexene. This selectivity is higher than for most of the
other trimerization processes. However, as all other trimeriza-
tion catalysts, the present one has the same disadvantage, i.e., the
production of 2–5 wt% of high molecular weight polyethylene.
Not only will this lead to lower yields, but the presence of high
molecular weight PE in the reaction mixture can cause reactor
fouling. This is especially true for the titanium-based process,
which currently runs at lower temperatures than the chromium-
based systems (30–80 ◦C versus 110–125 ◦C for the CPChem
process). It is, therefore, highly desirable to know which factors
are involved in the formation of polymeric products and how
their formation can be reduced or prevented.
We have identified that the PE formation turns out to be
catalyzed by at least two different species/processes. A signif-
icant amount is formed in early stage of the reaction and it is
most likely caused by the presence of partly alkylated titanium
species, which may act as polymerization catalysts. The PE for-
mation during later stages of the reaction is due to degraded
catalyst species, which means that polymer formation is an
inherent property of this catalyst system.
[3] (a) H. Ma, Y. Zhang, B. Chen, J. Huang, Y. Qian, J. Polym. Sci. Part
A: Polym. Chem. 39 (2001) 1817;
(b) D.H. Morgan, S.L. Schwikkard, J.T. Dixon, J.J. Nair, R. Hunter, R.
Adv. Synth. Catal. 345 (2003) 939;
(c) T. Monoi, Y. Sasaki, J. Mol. Catal. A: Chem. 187 (2002) 135;
(d) A. Carter, S.A. Cohen, N.A. Cooley, A. Murphy, J. Scutt, D.F. Wass,
Chem. Commun. (2002) 858;
(e) D.S. McGuinness, P. Wasserscheid, W. Keim, D. Morgan, J.T. Dixon,
A. Bollmann, H. Maumela, F. Hess, U. Englert, J. Am. Chem. Soc. 125
(2003) 5272;
(f) C. Andes, S.B. Harkins, S. Murtuza, K. Oyler, A. Sen, J. Am. Chem.
Soc. 123 (2001) 7423;
(g) Y. Chen, C. Qian, J. Sun, Organometallics 22 (2003) 1231;
(h) D.S. McGuinness, P. Wasserscheid, W. Keim, C. Hu, U. Englert, J.T.
Dixon, C. Grove, Chem. Commun. (2003) 334.
[4] (a) R.D. Knudsen, J.W. Freeman, US 2001053742 A1 (2001).;
(b) J.J.C. Grove, H.A. Mahomed, L. Griesel, WO 2003004158 A2 (2003)
to Sasol Tech Ltd.;
(c) M.G. Goode, T.E. Spriggs, I.J. Levine, W.R. Wilder, C.L. Edwards,
US 51,37,994 A (1992) to Union Carbide Chem. Plastics Techn.
Corp.
[5] W.M. Woodard, W.M. Ewert, H.D. Hensley, M.E. Lashier, B.E. Kreis-
cher, G.D. Cowan, J.W. Freeman, R.V. Franklin, R.D. Knudsen, R.L.
Anderson, L.R. Kallenbach, WO 99/19280 A1 (1999) to Phillips
Petroleum Co.
It has been shown that the PE formation in beginning of the
reaction can be greatly reduced by premixing the CpꢀTiCl3 and
MAO prior to injection into the reactor. This went hand in hand
with a significant increase in productivity. We also demonstrated
that the type of MAO activator/impurity scavenger is of great
importance. For a high productivity and a low yield of PE it is
essential to use MAOs that do not contain and/or are not able
to generate aluminum hydride species. The best results with
respect to both productivity and selectivity were obtained by
starting from trimethyl titanium compounds.
[6] K.J. Stone, K.J. Little, J. Org. Chem. 49 (1984) 1849.
[7] J.L.W. Pohlman, F.E. Brinkmann, Z. Naturforsch. 20b (1965) 5.
[8] (a) A.N.J. Blok, P.H.M. Budzelaar, A.W. Gal, Organometallics 22 (2003)
2564;
(b) T.J.M. de Bruin, L. Magna, P. Raybaud, H. Toulhoat,
Organometallics 22 (2003) 3404;
(c) S. Tobisch, T. Ziegler, Organometallics 22 (2003) 5392.
[9] A. Martin, M. Mena, M.A. Pellinghelli, P. Royo, R. Serrano, A. Tiripic-
chio, J. Chem. Soc., Dalton Trans. (1993) 2117.
[10] (a) S.W. Ewart, M.J. Sarsfield, E.F. Williams, M.C. Baird, J. Organomet.
Chem. 579 (1999) 106;
(b) S.W. Ewart, M.J. Sarsfield, D. Jeremic, T.L. Tremblay, E.F. Williams,
M.C. Baird, Organometallics 17 (1998) 1502.
[11] D.B. Malpass, Properties of Aluminoxanes from Akzo Nobel,
www.akzonobel-polymerchemicals.com.
Appendix A. Supplementary data
[12] W.P. Kretschmer, WO 2002070569 A1 (2002) to Stichting Dutch Poly-
mer Institute.
Supplementary data associated with this article can be found,
in the online version, at doi:10.1016/j.molcata.2005.12.030.
[13] The exact content of free AlMe3 in d-MAO is unknown. 1H NMR does
show the presence of free AlMe3, but after drying the material becomes
quite insoluble, making full characterization by NMR impossible. Nev-
ertheless, the content of free AlMe3 is expected to be significantly lower
than in PMAO.
References
[1] For an comprehensive overview of the field of ethylene trimerization
to 1-hexene see J.T. Dixon, M.J. Green, F.M. Hess, D.H. Morgan, J.
Organomet. Chem. 689 (2004) 3641.
[14] P. Arndt, A. Spannenberg, W. Baumann, S. Becke, U. Rosenthal, Eur.
J. Inorg. Chem. (2001) 2885.

H. Hagen et al. / Journal of Molecular Catalysis A: Chemical 248 (2006) 237–247
247
[15] C.N. Muhoro, J.F. Hartwig, Angew. Chem. Int. Ed. Engl. 36 (1997)
1510.
[16] G. Smith, S. Palmaka, J. Roger, D. Malpass, US 5,831,109 (1998) to
Akzo Nobel BV.
[17] (a) M.K. Mahanthappa, R.M. Waymouth, J. Am. Chem. Soc. 123 (2001)
12093;
[20] GC–MS analyses of the reaction mixtures showed (a) ENB: (min[M•+]):
10.62 [176], 10.66 [176], 10.73 [176], 10.76 [176]; 176 is the mass of
the ENB co-trimer which was found as mixture of four isomers, due to
the starting mixture of two isomers. (b) Norbornene (min[M•+]): 9.29
[150], 11.96 [206]; 150 is the mass of the norbornene co-trimer, one
isomer; 206 is the co-trimer of 150 and two molecules of ethylene. (c)
Norbornadiene (min[M•+]): 11.65 [204], 11.69 [204], 11.77 [204], 11.90
[204]; 204 is the mass of the norbornadiene bis co-trimer.
[21] (a) C. Pellechia, D. Pappalardo, G.-J. Gruter, Macromolecules 32 (1999)
4491;
(b) B. Ray, T.G. Neyroud, M. Kapon, Y. Eichen, M.S. Eisen,
Organometallics 20 (2001) 3044;
(c) Y.V. Kissin, A.J. Brandolini, J. Polym. Sci. Part A: Polym. Chem.
37 (1999) 4273.
[18] (a) D.A. Plattner, Int. J. Mass. Spectrom. 207 (2001) 125;
(b) A.P. Bruins, J. Chromatogr. A 794 (1998) 345.
[19] The loss of a hydrogen atom from the B-type complexes is probably
due to the ESI-MS process.
(b) C. Pellechia, D. Pappalardo, L. Oliva, M. Mazzeo, G.-J. Gruter,
Macromolecules 33 (2000) 2807.
[22] S. Murtuza, S.B. Harkins, G.S. Long, A. Sen, J. Am. Chem. Soc. 122
(2000) 1867.